Redundant Localization as the Foundation for Safe Autonomous Driving
Precise, robust, and fail-safe positioning is critical for autonomous driving. This is particularly important in rural environments, where infrastructure and surrounding conditions can vary and localization systems must operate reliably under real-world conditions.
Within the SUE project, Fraunhofer EMFT therefore developed a hybrid primary localization concept that intelligently combines several complementary approaches. The system integrates visual line detection for lane guidance with RFID-based reference points that are embedded along the route as infrastructural landmarks. This is complemented by a model-based state estimation that enables continuous calculation of the vehicle’s position.
The RFID tags serve as robust, weather-independent reference points that are read by the vehicle and compared with its internal position estimate. This additional reference layer creates a redundant system architecture capable of compensating for sensor failures and reducing uncertainties.
The different localization sources are combined using a Kalman-based sensor fusion approach. The result is a validated overall system that increases functional safety and enables stable vehicle positioning for autonomous driving at speeds of up to 50 km/h.